Extensive alteration in the expression profiles of TGFB ... · carcinoma (GC) is chronic atrophic...

Summary. Aims: The expression patterns of TGFBsignaling proteins, such as TGFB1/2, TGFBR1(ALK5),TGFBR2, SMAD1/2/3, SMAD2/3, SMAD4, SMAD7,and of downstream targets of TGFB signaling,CDKN1A (p21CIP1), CDKN1B (p27KIP1), MYC,CDC25A, TP53, and RELA (p65NF-kB) wereinvestigated in gastric carcinomas and other gastriclesions. Methods and results: A total of 112 gastriccarcinomas, 37 dysplasias, 54 intestinal metaplasias, 29chronic atrophic gastritis and 54 normal gastricepithelium were analyzed by tissue microarray-basedimmunohistochemical analysis. Extensive changes inexpression profiles of these proteins were observed.Three types of expression patterns were observed alongthe normal epithelium-atrophic gastritis-dysplasia-carcinoma sequence. (1) Expression of TGFB1/2,TGFBR1, MYC, and TP53 continually increased alongthis sequence. (2) Expression of SMAD4, CDKN1A,SMAD1/2/3, SMAD2/3, and CDKN1B was enhanced indysplasia but decreased in carcinoma. (3) Expression ofTGFBR2, SMAD7, RELA, and CDC25A was enhancedin dysplasia and the enhanced level was maintained incarcinoma. In addition, we also evaluated the clinicalsignificance of the expression of TGFB signalingproteins in gastric carcinoma. TGFB and MYC werepositively correlated with advanced stages, whereasSMAD1/2/3 and SMAD4 were strongly associated withearlier stages. Conclusions: The extensive change inexpression of TGFB signaling components is implicatedduring tumorigenesis of gastric neoplasias.

Key words: Gastric carcinoma, Gastric dysplasia,Intestinal metaplasia, TGFB, Immunohistochemistry

Introduction

The sequence of changes leading to gastriccarcinoma (GC) is chronic atrophic gastritis-dysplasia(intra-epithelial neoplasia)-carcinoma (Ho, 1996; Gentaand Rugge, 2001). Helicobacter pylori infection plays amajor role in the development of chronic atrophicgastritis, the first step of the cascade (Genta and Rugge,2001). The TGFB pathway occupies a central positionamong signaling networks that control growth,differentiation, apoptosis, and extracellular matrixaccumulation of a wide variety of cell types, as well asthe immune system. Despite its potent growth inhibitionin normal epithelial and lymphoid cells, it is evident thatTGFB signaling may act either as a suppressor or apromoter during carcinogenesis (Kim et al., 2000; Elliottand Blobe, 2005). However, a systematic analysis of theexpressional profile of the TGFB signaling pathwaycomponents during the sequence of changes leading tocarcinogenesis has not been conducted.

TGFB is a member of a large family of disulfide-bonded cytokines. Binding of TGFB to the TGFR2-TGFR1 (ALK5) heterodimer leads to phosphorylation ofTGFR1. TGFR1 activation results in phosphorylation ofparticular SMAD proteins, which form heteromericcomplexes and translocate to the nucleus, where theymay direct transcriptional responses (Derynck andZhang, 1996; Massague, 1996, 1998). SMAD2 and 3serve as receptor-regulated SMADs (R-SMADs)transducing TGFB/ACTIVIN-like signals, and SMAD1,5, and 8 act as R-SMADs, transducing BMP-like signals,while SMAD4 is the only common-partner SMAD (Co-SMAD) in mammals, and is shared by theTGFB/ACTIVIN and BMP pathways. SMAD6 andSMAD7 function as inhibitory SMADs (I-SMADs) in

Extensive alteration in the expression profiles of TGFBpathway signaling components and TP53 is observedalong the gastric dysplasia-carcinoma sequenceSeok-Hyung Kim1, Seung-Hyun Lee2, Yoon-La Choi1, Li-Hui Wang2,3, Cheol Keun Park1 and Young Kee Shin2

1Department of Pathology, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea, 2Laboratory of

Molecular Pathology, Department of Pharmacy, Seoul National University College of Pharmacy, Seoul, Korea and 3Department of

Pharmacology, Shenyang Pharmaceutical University College of Pharmacy, Shenyang, China

Histol Histopathol (2008) 23: 1439-1452

Offprint requests to: Young Kee Shin, M.D., Ph.D., Molecular PathologyLaboratory, Department of Pharmacy, Seoul National University Collegeof Pharmacy, San 56-1. Sillim-dong, Gwanak-gu, Seoul, 151-742,Korea. e-mail: [email protected]

http://www.hh.um.es

Histology andHistopathology

Cellular and Molecular Biology

Abbreviations: TGFB, transforming growth factor-ß; TBRI, transforminggrowth factor-ß receptor I; TBRII, transforming growth factor-ß receptor II

TGFB/ACTIVIN and BMP signaling. However,accumulating evidence suggests that TGFB/ACTIVINalso signal through other pathways, such as MAPK andNF-κB (de Guise et al., 2006) and that TP53 and NF-κBdirectly affect SMAD-dependent TGFB signaling(Cordenonsi et al., 2003, 2007). In the nucleus, R-SMAD-SMAD4 heterodimers interact with varioustranscription factors and transcriptional coactivators orcorepressors, leading to transcriptional regulation oftarget genes such as CDKN1A (p21CIP1), CDKN1B(p27KIP1), CDKN2B (p15INK4B), MYC, and CDC25A.

The expression of TGFB signaling proteins and itsclinical significance in GC have been examined inseveral studies. Increased expression of TGFB has beenreported in GC (Naef et al., 1997; Gold, 1999), whereasreduced expression of SMAD3 and 4 was recentlyreported to be correlated with unfavorable clinicalparameters in GC (Han et al., 2004; Kim et al., 2004;Wang et al., 2007). Overexpression of TP53 was alsocorrelated with a poor prognosis for patients (Okuyamaet al., 2002). However, these studies, for the most part,examined a single component of the TGFB pathway and,thus, there is some limitation in extrapolating themeaning of these results to the entire TGFB signalingpathway. Furthermore, the expression of TGFB signalingproteins and their downstream target proteins in gastricpre-cancerous lesions such as dysplasia have rarely beenstudied, with the exception of CDKN1A, so it is difficultto infer the potential roles during the early stages oftumorigenesis in the stomach. Therefore, a holistic andsystematic study of the expression of TGFB signalingproteins along the dysplasia-carcinoma sequence ofgastric neoplasias is required.

In this study, we employed tissue microarray-basedimmunohistochemical analysis as a large-scale screeningtechnique and studied the expression of all the TGFBsignaling proteins in GC and related lesions to provideinsights into potential therapeutic interventions thatcould be used to prevent and treat GC by modulating theTGFB signaling pathway.

Materials and methods

Patients, Samples and Tissue Microarrays

Tissues from patients with GC and other gastriclesions, including 112 GCs, 20 high grade dysplasias(HDs), 17 low grade dysplasias (LDs), 54 intestinalmetaplasias (IMs), 29 chronic atrophic gastritis (CAG)samples, and 54 normal gastric epithelium (NL)samples, were used for tissue microarray construction aspreviously described (Wang et al., 2007). This study wasreviewed and approved by the ethics committee of theinstitutional review board of Chungbuk NationalUniversity Hospital. The 112 gastric carcinomas [age =49-85 years; average age = 66.1 years; 40 female and 72male cases] were comprised of 30 early cases (pTis=1,pT1=29) and 82 advanced cases (pT2=26, pT3=42,pT4=14). A total of 25, 41, and 29 cases (26.5, 41.8, and

31.7%, respectively) were classified as well, moderately,and poorly differentiated adenocarcinomas, respectively.Histologically, 64 cases of GCs were intestinal type and31 cases of GCs were diffuse type. However, in allcases, IM or dysplasia was accompanied. All archivalmaterials were routinely fixed in 10% neutral-bufferedformalin and embedded in paraffin. These samples wereused to generate tissue microarray slides (3 mm indiameter) according to the standard method previouslydescribed (Choi et al., 2007; Wang et al., 2007).

Immunohistochemistry

The selection of the thirteen antibodies used in thisstudy was based on the known components of the TGFBsignaling pathway (Fig. 1), as well as the availability andsuitability of the antibody for paraffin-embeddedarchival materials (Table 1). The components examinedincluded TGFB signaling proteins [TGFB1/2, TGFBR1(ALK5), TGFBR2, SMAD1/2/3, SMAD2/3, SMAD4,SMAD7] and downstream targets of TGFB-regulatedtranscription [CDKN1A (p21CIP1), CDKN1B (p27KIP1),MYC, CDC25A], as well as RELA (p65NF-κB) andTP53. Once the antibodies were selected, the specificityof the primary antibodies used in these experiments wasconfirmed by Western blotting of nine human GC celllines, SNU5, SNU16, SNU216, SNU484, SNU638,MKN28, MKN74, KATOIII, and AGS (data not shown).Each antibody was then titrated with three to fivedifferent dilutions (with a 2-fold difference between eachdilution) on whole-mount tissue sections, according tothe manufacturer’s recommendations. If there were nowell-established positive control tissues, we used our in-house multi-tumor array to find a positive-control tissue.If signal-to-background ratio was not acceptable for thedilution tested, the antigen retrieval buffer, blockingsolution, incubation time, and antibody concentrationwere readjusted. The dilutions of primary antibody andthe antigen retrieval buffer used for each antibody arelisted in Table 1. Dextran polymer conjugated goat anti-mouse, or goat anti-rabbit antibodies and horseradishperoxidase (DAKO Glostrup Denmark, Envision+) wereused to detect primary antibody binding.Immunostaining was performed as previously described(Wang et al., 2007).

Evaluation of results of immunohistochemical staining

The evaluation of both the intensity ofimmunohistochemical staining and the proportion ofpositively-stained epithelial cells was performed aspreviously described (Wang et al., 2007). The stainingintensity was classified as: 1+, weak; 2+, moderate; or3+, strong. The percentage of positively staining tumorcells relative to the total number of tumor cells wasassigned to one of five ranges: 0, < 5%; 1, 5-25%; 2, 26-50%; 3, 51-75%; and 4, > 75%. The scores forpercentage of positive tumor cells and the stainingintensity were multiplied to produce a weighted

1440

TGFB pathway in gastric carcinogenesis

immunoreactive score (IS) for each tumor specimen.Each lesion was examined and scored separately by twopathologists, and cases with discrepant scores werereevaluated to achieve a consensus score. The expressionof TGFB1/2, TGFBR1, TGFBR2, SMAD7, MYC, andRELA in the cytoplasm was analyzed, while theexpression of SMAD1/2/3, SMAD2/3, and SMAD4 wasseparately analyzed in both cytoplasm and nucleus. Theexpression of CDKN1A, and CDKN1B, CDC25A, andTP53 was restricted to the nucleus (Fig. 2). In addition,the IS was classified into three groups: very weak (IS=0-1), moderate (2-5), and high (IS=6-12). Changes inexpression profile for each antigen were also analyzed.

Statistical analysis

Statistical analyses were performed using theFisher’s exact test, Pearson’s χ2 test, Mann-Whitney test,Kruskal-Wallis test, and the Tukey's HSD and Duncan'stests as post hoc tests. For comparison of means of ISs,the Mann-Whitney test, Kruskal-Wallis test, Tukey'sHSD, and Duncan's test were used. The association ofthe expression level with clinicopathological factors wasassessed by cross-tabulation, and significant differenceswere determined by the Fisher’s exact test and Pearson’sχ2 test. A P value less than 0.05 was regarded asstatistically significant. All statistical analyses wereperformed using SPSS software. (SPSS, Chicago, USA).

Results

The expression profiles of TGFB pathway signalingmolecules and TP53 in GC and related lesions

The mean IS of each molecule in GC and relatedlesions, including CAG, IM, and dysplasia are illustrated

in Fig. 3 (for detail see Table 2). Changes in expressionprofile were also similar when the staining levels weregrouped as very weak, moderate, and high intensitystaining (Table 3). These dynamic expression modes canbe categorized by three patterns (Fig. 4A). In the firstpattern, TGFB1/2, TGFBR1, MYC and TP53 showed acontinuous increase in expression along the NL CAG-dysplasia-GC sequence (Fig. 4A). All these moleculeswere significantly enhanced in GC (Tables 2, 3). Inaddition, TGFB and TGFBR1 expression was

1441


Table 1. Primary antibodies and antigen retrieval buffers used in this study.

Antibody Type Source Catologue No. Dilution Antigen retrieval buffer

TGFB1/2 Poly (R) S-cruz sc146 1:40 Borate-EDTATGFBR1 (ALK5) Poly (R) S-cruz sc398 1:40 Borate-EDTATGFBR2 E-6 (IgG2a) S-cruz sc17792 1:40 Borate-EDTASMAD2/3* Poly (G) S-cruz sc6032 1:40 Tris-EDTASMAD1/2/3 H-2 (IgG2a) S-cruz sc7960 1:50 Borate-EDTASMAD4 B-8 (IgG1) S-cruz sc7966 1:50 Borate-EDTASMAD7 Poly(R) S-cruz sc11392 1:30 Tris-EDTACDKN1A (p21CIP1) HZ52 (IgG1) Lab vision MS387 1:40 Borate-EDTACDKN1B (p27KIP1) F-8 (IgG1) S-cruz sc1641 1:30 Borate-EDTATP53 DO7 (IgG2b) Novocastra NCL-p53-DO7 1:50 CitrateRELA (p65NF-κB) F-6 (IgG1) S-cruz sc8008 1:50 Tris-EDTAMYC 9E11 (IgG1) DiNonA 55720E 1:50 CitrateCDC25A DCS120+/ DCS121 NeoMarker CDC25Ab-3 1:20 Borate-EDTA

Abbreviation: Poly(R), rabbit polyclonal, Poly(G), goat polyclonal, S-cruz, Santa Cruz Biotechnology, Borate-EDTA : 50 mM Borate (pH 8.2)supplemented with 1 mM EDTA and 1 mM NaCl, Tris-EDTA : 40 mM Tris (pH 9.5) supplemented with 1 mM EDTA, Citrate : 10 mM sodium citrate (pH6.0), SMAD2/3* : Rabbit anti-goat Antibody (Vector Labs, 1:250) was introduced after primary antibody incubation. Then Envision+ kit (anti-rabbit) wasapplied.

Fig. 1. Schematic drawing of the TGFB signaling pathway.

significantly elevated in dysplasia. In the second pattern,tumor suppressor proteins such as SMAD1/2/3,SMAD2/3, SMAD4, CDKN1A, and CDKN1B showedan inverse V-shaped pattern with a peak in dysplasia,which continuously increased from NL to dysplasia, butdecreased in GC (Fig. 4A). Expression of thesemolecules was significantly enhanced in dysplasia(Tables 2, 3). In the third pattern, TGFBR2, SMAD7,RELA, and CDC25A expression followed a sigmoidcurve type with a plateau from dysplasia to GC (Fig.4A). These proteins were upregulated in neoplasticlesions, including dysplasia and GC, but an additionalincrease in GC was not observed (Tables 2, 3).

Continuously increasing type: TGFB1/2, TGFBR1,MYC and TP53

In normal and non-neoplastic epithelia (CAG andIM), weak immunoreactivity for TGFB was rarely foundin the cytoplasm of the mucosal epithelial cells,particularly in the mucous neck region. However, TGFBimmunoreactivity was significantly enhanced in LD (IS:LD Vs NL, CAG P < 0.01 respectively) and reached itshighest level in GC (IS: CA Vs HD P < 0.025, CA VsLD P < 0.001, CA Vs non-neoplastic lesions P < 0.001).The overall expression pattern of TGFBR1 is similar tothat of TGFB. Immunoreactivity for TGFBR1 was alsorare and weak in non-neoplastic lesions, and was alsosignificantly elevated in dysplasia (IS: LD, HD Vs non-neoplastic epithelia P < 0.005 respectively). Expressionof TGFBR1 was most enhanced in carcinoma (IS: CAVs LD, HD P < 0.025, CA Vs non-neoplastic lesions P <0.001).

Immunoreactivities for MYC and TP53 werecharacteristically strong in GC compared to any otherprecursor lesions (IS: CA Vs all other lesions P < 0.01 inboth MYC and TP53), whereas the expression levels ofthese proteins were low and similar to each other in allnon-carcinoma lesions.

Inverse V-shaped type: SMAD1/2/3, SMAD2/3, SMAD4,CDKN1A, CDKN1B

Cytoplasmic SMAD1/2/3 (SMAD1/2/3(C))immunoreactivity was very rare and weak in NL, but its

level in preneoplastic lesions, including CAG and IMwere more elevated compared to NL (IS: NL Vs CAG P< 0.005, NL Vs IM P < 0.001). SMAD1/2/3(C) levelswere most highly enhanced in LD, but then weresignificantly reduced in HD and GC (IS: LD Vs anyother lesions P < 0.005). The overall expression patternof cytoplasmic SMAD2/3 (SMAD2/3(C)) was verysimilar to that of SMAD1/2/3. However, the highestexpression of SMAD2/3(C) was observed in HD, not inLD, and significant down-regulation in GC wasdisclosed (IS: HD Vs CA P < 0.05) The overallexpression pattern of cytoplasmic SMAD4 (SMAD4(C))was also similar to SMAD1/2/3(C) and SMAD2/3(C),with minor differences. Its level was highest in LD andabruptly reduced in GC (IS: LD, HD Vs CA P < 0.05).In contrast to the cytoplasmic expression of SMAD1/2/3,the immunoreactivity of nuclear SMAD1/2/3(SMAD1/2/3(N)) was frequently observed in NL, andthere was no significant difference in expressional levelamong all lesions. However, the nuclear staining ofSMAD2/3 (SMAD 2/3(N)) was strongest in HD and alsosignificantly decreased in GC like SMAD2/3(C) (IS: HDVs all other lesions P < 0.01). The expression pattern innuclear SMAD4 (SMAD4(N)) was similar toSMAD4(C) pattern and the highest expression was notedin dysplasia, but fell dramatically in carcinoma, and wasreduced to the level of non-neoplastic lesions (IS: LD VsCA P < 0.005, HD Vs CA P < 0.001).

CDKN1A (p21) was expressed weakly in thenucleus of epithelial cells of normal gastric mucosa andits expression level was enhanced in non-neoplasticinflammatory lesions, including CAG and IM (IS: NLVs CAG, IM P < 0.001). The strongestimmunoreactivity was observed in the nucleus of tumorcells of both low and high grade dysplasia (IS: LD, HDVs NL, CAG, IM P < 0.001 in all combinations).However, levels were reduced in carcinoma (IS: CA VsLD, HD P < 0.001). CDKN1B (p27) was also expressedin the nucleus of various cells, and levels in normal andnon-neoplastic lesions were very low. Similar to othertumor suppressor proteins, levels of CDKN1B (p27)were greatly increased in dysplasia (IS: LD, HD Vs NL,CAG, IM P < 0.001 in all combinations). The level wasalso markedly down-regulated in carcinoma (IS: CA VsLD, HD P < 0.001).

1442


Table 2. The immunoreactivity scores of TGFB signaling components in carcinoma, dysplasia, metaplasia, chronic atrophic gastritis and normalepithelium of the stomach.

TGFB 1/2 TGFBR1 TGFBR2 SMAD 1/2/3 SMAD 2/3 SMAD 4 SMAD7 CDKN1A CDKN1B MYC CDC25A REL A TP53

(ALK5) Cyt Nu Cyt Nu Cyt Nu (p21) (p27) (p65NFKB)

CA 4.04±2.44 5.23±2.93 2.89±1.88 3.27±2.21 1.72±1.98 2.68±1.89 2.32±2.12 2.46±2.12 2.12±2.18 4.72±2.82 5.10±3.29 1.49±1.74 4.90±2.71 4.99±2.75 5.85±3.25 2.67±3.89

AD (HD) 2.20±1.15 4.00±2.22 2.10± 1.37 3.10±1.92 2.00±1.75 4.33±2.14 3.83±2.31 4.60±1.31 3.55±1.70 4.45±2.39 8.40±3.69 3.85±2.96 2.39±1.58 4.94±3.30 4.89±2.21 0.84±2.85

AD (LD) 2.76±1.39 3.88±1.83 3.12± 1.22 4.76±1.09 1.82±1.59 2.21±1.67 1.71±1.44 5.59±2.53 3.47±1.84 5.53±1.50 8.29±3.18 4.00±2.88 3.08±2.60 5.64±1.78 5.53±2.00 0.15±0.55

IM 1.71±1.55 2.31±1.45 1.46± 1.16 2.29±1.50 1.45±1.29 2.00±1.48 2.88±1.35 2.86±1.57 2.49±1.43 3.30±2.27 3.45±2.49 0.95±0.80 3.18±1.77 3.16±2.05 2.71±1.58 0.00±0.00

CAG 1.04±1.18 1.67±1.52 1.68± 1.36 2.03±1.70 2.03±1.21 1.13±1.15 2.38±1.17 2.00±1.86 2.79±1.62 2.52±2.16 4.35±2.39 0.80±0.71 2.54±1.84 3.61±2.11 1.72±1.57 0.00±0.00

NL 1.31±1.85 0.89±0.98 0.96± 1.16 0.86±1.02 1.80±1.30 0.15±0.44 2.12±1.23 0.49±0.78 1.57±1.38 0.92±1.05 1.40±1.81 0.13±0.33 2.08±1.86 3.08±1.86 0.85±0.92 0.00±0.00

1443


Fig. 2. Immunostaining of gastric carcinomas and related lesions including dysplasia (high/lowgrade), intestinal metaplasia, chronic atrophic gastritis, and normal epithelial tissue using antibodiesrecognizing each TGFB signaling component and TP53. Abbreviations: CA, carcinoma; HD, Highgrade dysplasia; LD, low grade dysplasia; IM, intestinal metaplasia; CAG, Chronic atrophic gastritis;NL, normal gastric epithelium.

1444


Table 3. The expression profiles of TGFB signaling components and TP53 in gastric carcinomas and related precursor lesions.

NL CAG IM LD HD CA P-value

TGFB1/2

W* 35/52 (67.3%) 18/26 (69.2%) 20/42 (47.6%) 2/17 (11.8%) 4/20 (20.0%) 20/110 (18.2%)P<0.001 (LD vs IM: P<0.001,CA vs HD: P<0.025)

M** 14/52 (26.9%) 8/26 (30.8%) 21/42 (50.0%) 15/17 (88.2%) 16/20 (80.0%) 57/110 (51.8%)

S*** 3/52 (5.8%) 0/26 (0.0%) 1/42 (2.4%) 0/17 (0.0%) 0/20 (0.0%) 33/110 (30.0%)

TGFBR1

W 42/54 (77.8%) 16/27 (59.3%) 16/51 (31.4%) 2/17 (11.8%) 4/20 (20.0%) 16/109 (14.7%) P<0.001 (IM vs CAG: P<0.025,LD vs IM: P<0.001, CA vs LD:P<0.025)

M 12/54 (22.2%) 10/27 (37.0%) 35/51 (68.6%) 12/17 (70.6%) 11/20 (55.0%) 39/109 (35.8%)

S 0/54 (0.0%) 1/27 (3.7%) 0/51 (0.0%) 3/17 (17.6%) 5/20 (25.0%) 54/109 (49.5%)

TGFBR2

W 39/53 (73.6%) 13/28 (46.4%) 33/54 (61.1%) 1/17 (5.9%) 9/20 (45.0%) 31/109 (28.4%) P<0.001 (CAG vs NL: P<0.025,LD vs IM: P<0.001, HD vs LD:P<0.025)

M 14/53 (26.4%) 13/28 (46.4%) 21/54 (38.9%) 15/17 (88.2%) 11/20 (55.0%) 55/109 (50.5%)

S 0/53 (0.0%) 2/28 (7.1%) 0/54 (0.0%) 1/17 (5.9%) 0/20 (0.0%) 23/109 (21.1%)

SMAD1/2/3(Cytoplasm)

W 41/51 (80.4%) 12/29 (41.4%) 21/55 (38.2%) 0/17 (0.0%) 5/20 (25.0%) 30/111 (27.0%) P<0.001 (CAG vs NL: P<0.001,LD vs IM: P<0.001, CA vs LD:P<0.05)

M 10/51 (19.6%) 15/29 (51.7%) 33/55 (60.0%) 14/17 (82.4%) 14/20 (70.0%) 68/111 (61.3%)

S 0/51 (0.0%) 2/29 (6.9%) 1/55 (1.8%) 3/17 (17.6%) 1/20 (5.0%) 13/111 (11.7%)

SMAD1/2/3(Nucleus)

W 23/51 (45.1%) 8/29 (27.6%) 32/55 (58.2%) 10/17 (58.8%) 9/20 (45.0%) 62/111 (55.9%)P<0.05 (IM vs CAG: P<0.01)

M 28/51 (54.9%) 21/29 (72.4%) 23/55 (41.8%) 7/17 (41.2%) 11/20 (55.0%) 44/111 (39.6%)

S 0/51 (0.0%) 0/29 (0.0%) 0/55 (0.0%) 0/17 (0.0%) 0/20 (0.0%) 5/111 (4.5%)

SMAD2/3(Cytoplasm)

W 33/34 (97.1%) 18/24 (75.0%) 16/41 (39.0%) 5/14 (35.7%) 3/18 (16.7%) 34/109 (31.2%) P<0.001 (CAG vs NL: P<0.025,IM vs CAG: P<0.01, HD vs LD:P<0.025, CA vs HD: P<0.01)

M 1/34 (2.9%) 6/24 (25.0%) 25/41 (61.0%) 9/14 (64.3%) 9/18 (50.0%) 68/109 (62.4%)

S 0/34 (0.0%) 0/24 (0.0%) 0/41 (0.0%) 0/14 (0.0%) 6/18 (33.3%) 7/109 (6.4%)

SMAD2/3(Nucleus)

W 10/34 (29.4%) 6/24 (25.0%) 5/41 (12.2%) 6/14 (42.9%) 3/18 (16.7%) 46/109 (42.2%)P<0.001 (LD vs IM: P<0.05, CAvs IM: P<0.001)

M 24/34 (70.6%) 18/24 (75.0%) 34/41 (82.9%) 8/14 (57.1%) 11/18 (61.1%) 53/109 (48.6%)

S 0/34 (0.0%) 0/24 (0.0%) 2/41 (4.9%) 0/14 (0.0%) 4/18 (22.2%) 10/109 (9.2%)

SMAD4(Cytoplasm)

W 46/53 (86.8%) 13/27 (48.1%) 11/49 (22.4%) 1/17 (5.9%) 0/20 (0.0%) 42/112 (37.5%) P<0.001 (CAG vs NL: P<0.001,LD vs IM: P<0.001, HD vs LD:P<0.05, CA vs HD: P<0.001)

M 7/53 (13.2%) 14/27 (51.9%) 37/49 (75.5%) 9/17 (52.9%) 18/20 (90.0%) 64/112 (57.1%)

S 0/53 (0.0%) 0/27 (0.0%) 1/49 (2.0%) 7/17 (41.2%) 2/20 (10.0%) 6/112 (5.4%)

SMAD4(Nucleus)

W 26/53 (49.1%) 6/28 (21.4%) 12/49 (24.5%) 3/17 (17.6%) 2/20 (10.0%) 57/112 (50.9%)P<0.001 (CAG vs NL: P<0.025,CA vs HD: P<0.01)

M 27/53 (50.9%) 20/28 (71.4%) 36/49 (73.5%) 12/17 (70.6%) 14/20 (70.0%) 48/112 (42.9%)

S 0/53 (0.0%) 2/28 (7.1%) 1/49 (2.0%) 2/17 (11.8%) 4/20 (20.0%) 7/112 (6.3%)

SMAD7

W 40/48 (83.3%) 10/25 (40.0%) 12/50 (24.0%) 0/17 (0.0%) 1/20 (5.0%) 15/110 (13.6%)P<0.001 (CAG vs NL: P<0.001,LD vs IM: P<0.001)

M 8/48 (16.7%) 12/25 (48.0%) 29/50 (58.0%) 8/17 (47.1%) 11/20 (55.0%) 54/110 (49.1%)

S 0/48 (0.0%) 3/25 (12.0%) 9/50 (18.0%) 9/17 (52.9%) 8/20 (40.0%) 41/110 (37.3%)

CDKN1A

W 35/53 (66.0%) 1/23 (4.3%) 10/47 (21.3%) 0/17 (0.0%) 1/20 (5.0%) 22/109 (20.2%) P<0.001 (CAG vs NL: P<0.001,LD vs IM: P<0.001, CA vs HD:P<0.05)

M 15/53 (28.3%) 16/23 (69.6%) 28/47 (59.6%) 3/17 (17.6%) 4/20 (20.0%) 39/109 (35.8%)

S 3/53 (5.7%) 6/23 (26.1%) 9/47 (19.1%) 14/17 (82.4%) 15/20 (75.0%) 48/109 (44.0%)

CDKN1B

W 48/48 (100.0%) 21/25 (84.0%) 31/41 (75.6%) 3/14 (21.4%) 7/20 (35.0%) 66/110 (60.0%) P<0.001 (CAG vs NL: P<0.01,LD vs IM: P<0.001, CA vs HD:P<0.001)

M 0/48 (0.0%) 4/25 (16.0%) 10/41 (24.4%) 7/14 (50.0%) 6/20 (30.0%) 41/110 (37.3%)

S 0/48 (0.0%) 0/25 (0.0%) 0/41 (0.0%) 4/14 (28.6%) 7/20 (35.0%) 3/110 (2.7%)

MYC

W 21/39 (53.8%) 9/24 (37.5%) 8/38 (21.1%) 4/13 (30.8%) 7/18 (38.9%) 12/111 (10.8%)

P<0.001 (CA vs HD: P<0.01)M 14/39 (35.9%) 14/24 (58.3%) 26/38 (68.4%) 6/13 (46.2%) 10/18 (55.6%) 54/111 (48.6%)

S 4/39 (10.3%) 1/24 (4.2%) 4/38 (10.5%) 3/13 (23.1%) 1/18 (5.6%) 45/111 (40.5%)

CDC25AC(Nucleus)

W 9/39 (23.1%) 4/28 (14.3%) 8/43 (18.6%) 0/14 (0.0%) 1/18 (5.6%) 8/108 (7.4%)

P<0.001 (LD vs IM: P<0.01)M 27/39 (69.2%) 19/28 (67.9%) 29/43 (67.4%) 6/14 (42.9%) 10/18 (55.6%) 56/108 (51.9%)

S 3/39 (7.7%) 5/28 (17.9%) 6/43 (14.0%) 8/14 (57.1%) 7/18 (38.9%) 44/108 (40.7%)

RELA

W 36/46 (78.3%) 15/25 (60.0%) 9/41 (22.0%) 0/17 (0.0%) 1/19 (5.3%) 15/110 (13.6%)P<0.001 (IM vs CAG: p<0.01,LD vs IM: P<0.01)

M 10/46 (21.7%) 8/25 (32.0%) 28/41 (68.3%) 9/17 (52.9%) 9/19 (47.4%) 35/110 (31.8%)

S 0/46 (0.0%) 2/25 (8.0%) 4/41 (9.8%) 8/17 (47.1%) 9/19 (47.4%) 60/110 (54.5%)

TP53

W 62/62 (100.0%) 26/26 (100.0%) 54/54 (100.0%) 12/13 (92.3%) 17/19 (89.5%) 69/111 (62.2%)P<0.005 (CA vs IM/CAG/NL:P<0.001)

M 0/62 (0.0%) 0/26 (0.0%) 0/54 (0.0%) 1/13 (7.7%) 1/19 (5.3%) 13/111 (11.7%)

S 0/62 (0.0%) 0/26 (0.0%) 0/54 (0.0%) 0/13 (0.0%) 1/19 (5.3%) 29/111 (26.1%)

*Very weak (W): Immunoreactive Score (IS) = 0-1; **Moderate (M): IS = 2-5; ***Strong (S): IS = 6-12. Abbreviations: CA, Carcinoma; HD, high gradedysplasia; LD, Low grade dysplasia; IM, Intestinal metaplasia; CAG, Chronic atrophic gastritis; NL, Normal.

Sigmoid curve type: TGFBR2, SMAD7, RELA, CDC25A

Immunoreactivity for TGFBR2 was weak in non-neoplastic epithelium, especially in normal epithelium(IS: NL Vs CAG P < 0.05), but the level wassignificantly enhanced in dysplasia and carcinoma (IS:LD, CA Vs any non-neoplastic epithelium P < 0.005).However, in contrast to TGFB, TGFBR1, MYC, andTP53, the level of TGFBR2 expression appeared toplateau in dysplasia and did not increase further in

carcinoma. Similar expression patterns were observedfor SMAD7, RELA (NF-κB), CDC25A.

Clinicopathological correlation of the expression of TGFBsignaling proteins and TP53 in GC

The correlation of TGFB signaling protein and TP53expression in gastric adenocarcinomas with variousclinicopathological parameters was analyzed. TGFB andMYC were significantly correlated with advanced

1445


Fig. 3. The expression profile of each TGFBpathway component and TP53 in tissue samplesof gastric cancers and related lesions. Themean immunoreactive score (IS) for eachcomponent in each type of lesion is showngraphically. The TGFB pathway componentscan be divided into TGFB and its receptors (A),R-SMADs including SMAD1/2/3 and SMAD2/3(B), Co-SMADs and I-SMADs, such as SMAD4and SMAD7 (C), tumor suppressive effectors,including CDKN1A, CDKN1B and TP53 (D), pro-oncogenic effectors, including MYC, CDC25A,RELA (E). Abbreviations: CA, carcinoma; HD,High grade dysplasia; LD, low grade dysplasia;IM, intestinal metaplasia; CAG, chronic atrophicgastritis; NL, normal gastric epithelium.

Table 4. Correlation of clinicopathologic parameters and immunoreactivity scores of TGF pathway components in gastric carcinoma.

TGFB1/2 TGFBR1 TGFBR2 SMAD1/2/3 SMAD1/2/3 SMAD2/3 SMAD2/3 SMAD4 SMAD4 SMAD7 CDKN1A CDKN1B MYC CDC25A RELA TP53

(C) (N) (C) (N) (C) (N)

Age (No.) N.S. P=0.06 P=0.10 P=0.15 N.S N.S N.S N.S N.S P<0.05 P<0.05 P<0.05 N.S N.S N.S P<0.05

≤ 55 years (32) 3.87±2.28 4.31±2.87 2.40±1.61 2.78±2.06 1.56±1.85 2.27±1.78 2.37±2.13 2.09±1.99 1.91±1.86 3.84±2.67 4.06±2.95 0.94±1.29 4.42±3.11 4.63±2.63 5.48±3.33 1.42±3.16

> 55years (79) 4.10±2.53 5.53±2.91 3.08±1.95 3.44±2.26 1.78±2.05 2.84±1.92 2.30±2.13 2.62±2.17 2.20±2.31 5.06±2.82 5.47±3.37 1.71±1.85 5.09±2.54 5.13±2.80 6.00±3.23 3.15±4.05

Sex (No.) N.S. N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S N.S

Female (40) 4.18±2.60 5.03±2.80 2.84±1.91 3.15±2.43 1.90±2.24 2.42±1.84 2.53±2.23 2.13±2.29 2.10±2.18 4.77±2.52 5.26±3.45 1.37±1.63 4.61±2.83 5.14±2.80 5.87±3.16 3.03±3.89

Male (72) 3.96±2.39 5.30±3.02 2.92±1.88 3.31±2.09 1.62±1.85 2.82±1.91 2.21±2.07 2.65±2.02 2.13±2.19 4.69±2.99 4.94±3.23 1.56±1.80 5.05±2.65 4.92±2.74 5.85±3.32 2.48±3.90

LN metastasis (No.) P=0.06 N.S N.S N.S P<0.05 N.S N.S P<0.01 P<0.01 N.S N.S N.S P=0.127 P<0.05 N.S N.S

Negative (41) 3.46±2.13 5.32±3.38 2.66±1.78 3.39±1.93 2.17±2.00 2.88±1.82 2.43±2.06 3.29±2.24 2.95±2.19 4.83±2.50 5.37±3.11 1.59±1.87 4.40±2.59 4.24±2.02 5.73±3.05 2.48±3.87

Positive (68) 4.38±2.60 5.13±2.67 3.00±1.94 3.14±2.38 1.48±1.96 2.56±1.95 2.29±2.17 1.96±1.91 1.63±2.05 4.67±3.02 4.76±3.33 1.46±1.67 5.22±2.77 5.38±3.01 5.90±3.40 2.82±3.94

Tumor size (No.) N.S P=0.12 N.S N.S P<0.01 N.S N.S P=0.17 P<0.01 P<0.05 N.S N.S P<0.05 P<0.05 P<0.05 N.S

≥ 5.0cm (65) 4.30±2.74 5.64±2.72 3.11±2.00 3.39±2.35 1.11±1.72 2.81±1.86 2.24±2.27 2.19±2.07 1.56±2.09 5.28±3.10 4.89±3.12 1.30±1.50 5.46±2.68 5.59±3.08 6.50±3.07 2.84±3.97

< 5.0cm (45) 3.76±2.10 4.76±3.10 2.67±1.75 3.11±2.08 2.35±2.07 2.55±1.92 2.40±1.98 2.75±2.16 2.69±2.14 4.11±2.36 5.22±3.49 1.69±1.95 4.33±2.65 4.39±2.24 5.19±3.32 2.49±3.83

Gross type (No.) P<0.01 P<0.05 P<0.05 N.S P<0.001 N.S P<0.05 P<0.05 P<0.001 N.S P=0.18 N.S P<0.01 P<0.05 P=0.07 N.S

EGC (30) 2.90±1.83 4.33±3.30 2.27±1.78 2.87±1.87 3.13±2.11 2.97±1.99 2.81±1.80 3.23±1.96 3.80±2.12 4.17±2.28 5.73±3.51 1.48±1.41 3.52±2.69 4.07±2.10 4.93±2.86 2.10±3.52

AGC (82) 4.46±2.53 5.53±2.73 3.13±1.88 3.40±2.32 1.20±1.68 2.56±1.85 2.13±2.22 2.18±2.13 1.50±1.86 4.91±2.98 4.80±3.20 1.49±1.86 5.44±2.54 5.35±2.90 6.20±3.34 2.89±4.02

Histological Type (No.) P<0.05 P<0.01 P<0.01 P<0.001 N.S P<0.01 P<0.05 P<0.001 P=0.12 P<0.05 P=0.12 P<0.001 P<0.01 P<0.05 P<0.05 P<0.05

Intestinal (64) 4.69±2.34 6.19±2.61 3.55±1.66 4.08±1.92 1.69±2.06 3.31±1.77 2.62±2.33 3.17±2.15 2.28±2.35 5.52±2.70 5.68±3.25 2.06±1.89 5.78±2.31 5.78±2.75 6.67±2.78 3.49±4.22

Diffuse (31) 3.48±2.35 4.50±2.69 2.33±1.86 2.41±2.20 1.59±1.95 1.77±1.52 1.67±1.52 1.48±1.54 1.58±1.64 4.03±2.66 4.55±3.32 0.80±1.10 4.13±2.55 4.07±2.12 5.32±3.17 1.52±3.06

Grade (No.) P<0.05 P<0.05 P<0.05 P<0.05 N.S P<0.01 P=0.06 P<0.05 P<0.05 P<0.05 P<0.05 P<0.01 P<0.01 N.S P<0.05 P=0.08

I (25) 4.60±2.31 6.48±2.96 3.32±1.60 4.20±1.85 2.00±1.71 3.81±1.58 2.96±2.14 4.16±1.57 3.40±2.04 5.28±2.81 7.13±2.51 2.65±2.08 4.88±1.88 5.56±2.60 6.44±2.75 2.92±4.19

II (41) 4.90±2.37 6.22±2.40 3.85±1.64 4.17±2.02 1.56±2.20 3.02±1.77 2.49±2.49 2.68±2.29 1.66±2.42 6.05±2.69 5.20±3.39 1.71±1.65 6.27±2.31 5.78±2.95 7.24±2.89 3.83±4.21

III (29) 3.31±2.33 4.32±2.54 2.07±1.68 2.17±1.91 1.60±2.01 1.82±1.68 1.57±1.50 1.35±1.36 1.52±1.67 3.63±2.20 4.03±3.01 0.79±1.10 4.48±2.89 4.37±1.90 5.14±3.10 1.90±3.34

Tumor invasion (No.) P<0.05 P<0.05 P=0.10 P=0.10 P<0.001 P=0.08 N.S P=0.07 P<0.001 P=0.08 P<0.05 N.S P<0.01 P=0.09 P=0.08 N.S

I (29) 2.93±1.85 4.24±3.32 2.24±1.81 2.83±1.89 2.97±1.94 2.97±2.03 2.90±1.75 3.24±1.99 3.69±2.07 4.25±2.29 5.52±3.37 1.40±1.35 3.60±2.70 4.10±2.13 4.97±2.91 2.07±3.57

II (26) 4.46±2.40 6.32±2.79 3.36±1.50 3.62±2.17 2.15±2.22 3.08±1.35 2.00±2.47 2.41±2.42 1.74±2.12 4.81±2.50 5.84±3.40 1.80±2.57 4.92±2.62 4.80±2.20 6.44±3.10 3.38±4.45

III (42) 4.35±2.82 4.95±2.78 2.87±2.19 2.90±2.46 0.73±1.32 2.08±2.07 2.26±2.16 2.02±1.98 1.54±1.94 4.44±3.22 3.83±3.08 1.55±1.43 5.50±2.61 5.23±3.19 5.61±3.69 2.83±3.78

IV (14) 4.77±2.13 6.15±2.12 3.46±1.33 4.38±1.94 1.23±1.83 3.23±1.64 2.08±2.10 2.31±1.97 1.38±1.80 6.54±2.90 5.92±2.35 1.15±1.57 6.15±2.44 6.31±2.95 7.54±2.18 2.54±4.12

Stage (No.) P<0.05 N.S N.S N.S P<0.001 N.S N.S P<0.05 P<0.001 N.S N.S N.S P<0.05 N.S N.S N.S

I (39) 3.21±2.04 4.77±3.35 2.49±1.75 3.10±1.94 2.74±2.11 3.00±1.84 2.70±1.99 3.28±2.26 3.38±2.21 4.34±2.28 5.77±3.46 1.67±1.96 3.95±2.68 4.26±2.02 5.33±3.02 2.13±3.67

II (10) 4.80±1.99 5.22±0.97 3.22±1.56 3.70±2.21 1.70±2.16 3.00±1.66 2.44±2.40 2.27±2.28 2.00±1.95 4.10±1.52 4.50±3.14 1.30±1.77 5.30±3.40 5.33±2.45 6.56±2.88 2.40±3.60

III (35) 4.69±2.74 5.66±3.06 3.11±2.00 3.28±2.54 0.83±1.40 2.47±1.96 2.09±2.37 2.11±1.86 1.47±2.05 4.94±3.27 4.62±3.18 1.60±1.63 5.40±2.63 5.03±2.95 5.94±3.79 2.97±4.11

IV (25) 4.12±2.57 5.24±2.60 3.00±2.02 3.20±2.24 1.48±1.85 2.36±2.00 2.08±1.89 1.72±1.90 1.12±1.51 5.24±3.29 4.48±2.97 1.20±1.55 5.60±2.33 5.79±3.28 6.20±2.97 3.32±4.15

Fig. 4. A. Three types of expression patterns were observed for TGFB signaling molecules: consistently increasing (a), inverse V-shaped (b), andsigmoid curve (c). B. Immunostaining of gastric carcinoma tissues using antibodies which recognize each of the TGFB signaling molecules and TP53.Prominent cytoplasmic staining (TGFB, TGFBR1, TGFBR2, SMAD7, RELA and MYC), or nuclear staining (CDKN1A and TP53), or both cytoplasmicand nuclear staining (SMAD1/2/3, SMAD2/3, SMAD4, CDKN1B and CDC25A) were observed. Abbreviations: CA, carcinoma; HD, high gradedysplasia; LD, low grade dysplasia; IM, intestinal metaplasia; CAG, chronic atrophic gastritis; NL, normal gastric epithelium.

clinical stages, whereas SMAD1/2/3(N), andSMAD4(C) and SMAD4(N), were correlated with earlyclinical stages (Table 4). TGFB, TGFBR1, and MYCwere significantly enhanced in cases of deep tumorinvasion, whereas SMAD1/2/3(N) and SMAD4(N) wereelevated in cases of superficial invasion (Table 4).TGFB, TGFRI, TGFRII, MYC, and nuclear CDC25Awere more highly expressed in AGC (advanced gastriccancer) than in EGC (early gastric cancer). In contrast,SMAD1/2/3(N) and SMAD2/3(N), and SMAD4(C) andSMAD4(N) were significantly upregulated in EGC(Table 4). CDC25A was positively correlated with nodalmetastasis, while SMAD1/2/3(N), and SMAD4(C) andSMAD4(N) were inversely correlated with nodalmetastasis (Table 4). The degree of SMAD7, MYC,

1447


Fig. 6. Schematic drawing of the expression profiles ofTGFB signaling components and TP53 during the chronicatrophic gastrit is-dysplasia-carcinoma sequence.Abbreviations: CA, carcinoma; AD, Adenoma; HD, Highgrade dysplasia; LD, low grade dysplasia; IM, intestinalmetaplasia; CAG, Chronic atrophic gastritis; NL, normalgastric epithelium.

Fig. 5. Schematic presentation of a hypothesis for the effects ofalterations in TGFB signaling during tumorigenesis.

CDC25A, and RELA up-regulation was significantlycorrelated with increased tumor size, whereasSMAD1/2/3(N) and SMAD4(N) were inverselycorrelated with tumor size (Table 4).

Discussion

In this study, we have demonstrated dynamicchanges in expression of TGFB signaling componentsand TP53 in gastric adenocarcinoma and related lesions.Based on our findings, our hypothesis regarding thechange in character of TGFB signaling during thetumorigenesis of gastric neoplasms is illustrated in Fig.5.

CAG, metaplasia, and dysplasia

In CAG, we noted increased expression of both pro-oncogenic (SMAD7, RELA) and tumor suppressorproteins, including SMAD1/2/3(C), SMAD2/3(C)SMAD4(C/N), CDKN1A, and CDKN1B (Fig. 6). Indysplasia, more extensive changes in the expressionprofile were observed, and they are also characterized byelevated expression of both pro-oncogenic (TGFB,TGFBR1, TGFBR2) and tumor suppressor proteins,including SMAD7, RELA, SMAD1/2/3, SMAD4,

CDKN1A, and CDKN1B. In CAG and dysplasia,TGFB1 signaling possibly modulates the directinduction of CDKN1A gene transcription (Pardali et al.,2000) and the regulation of CDKN1B (Sandhu et al.,1997) via a SMAD-dependent pathway. In addition,SMAD7 up-regulation in CAG and dysplasia could berelated to the induction of transient expression ofSMAD7 by negative feedback modulation of a TGFB1signal (Li et al., 2005). In CAG and dysplasia, epitheliamay try to counteract various cell proliferation stimuliby activation of TGFB1 signaling. Moreover, theenhanced expression of TGFB and its receptors mayactivate pro-oncogenic, SMAD-independent pathways,particularly Ras/MAPK/MEK signaling, as well astumor suppressive, SMAD-dependent pathways(Wakefield and Roberts, 2002) (Fig. 5). Therefore, theoverall effects of TGFB signaling in CAG and dysplasiamay be unaltered due to the balance between pro-oncogenic and tumor suppressor activities (Fig. 5).

Comparison of our results with previous studies of theexpression of TGFB signaling proteins in GC

Our results are generally well consistent withprevious studies. The high expression of TGFB in AGChas been reported in previous studies (Kai et al., 1996;

1448


Table 5. Expression patterns of TGFB signaling molecules and TP53 in gastric carcinoma (GC) that have been reported in the literature.

Expression genetic/epigenetic alterations Our data#

TGFB1/223-79% positive in GC(Maehara et al., 1999; Saito et al., 1999; Ebert et al., 2000;Xiangming et al., 2001; Zolota et al., 2002)

— 30.0% strong positive in GC

TGFBR1 82% downregulation in GC (Ito et al., 1992)13-64% methylation (Kang et al., 1999;Pinto et al., 2003)

49.5% strong positive in GC

TGFBR2 42% downregulation in GC (Takeno et al., 2002)33%LOH 0-74% mutation (Guo et al.,1998; Takeno et al., 2002)


SMAD2/3 38% downregulation in GC (SMAD3) (Han et al., 2004)no mutation (SMAD2) (Shitara et al.,1999)

42.2% downregulation in GC*

SMAD413-75% downregulation in GC(Xiangming et al., 2001; Kim et al., 2004)

10% mutation, 29-45% LOH, 5%methylation (Wang et al., 2007)


SMAD7 32% positive in GC (Kim et al., 2004) 37.3% strong positive in GC

CDKN1A30-62% downregulation in GC (Jang et al., 1998; Park et al., 1998;Xie et al., 2004; Al-Moundhri et al., 2005)

no mutation Park et al., 1998 27%polymorphism (Xie et al., 2004)


CDKN1B52-60% downregulation in GC(Jang et al., 1998; Al-Moundhri et al., 2005)

no mutation (Shin et al., 2000) 60.0% downregulation in GC*

MYC47-69% overexpression in GC(Spandidos et al., 1991; Onoda et al., 1996; Hara et al., 1998; Hanet al., 1999; Kozma et al., 2001)

Amplification (Hara et al., 1998;Kozma et al., 2001)


CDC25A38% overexpression in GC(Kudo et al., 1997)


RELA70% positive in GC(Sasaki et al., 2001)


TP5350% overexpression in GC (Fenoglio-Preiser et al., 2003)

37%LOH 34%mutation(Fenoglio-Preiser et al., 2003)


# obtained from Supplementary Table 2, * Nuclear staining.

Naef et al., 1997) and has been positively correlated withinvasion and metastasis, in accordance with previousstudies (Maehara et al., 1999; Saito et al., 1999) (Table5). There are very few reports which have examinedTGFBR expression in GC tissues. According to Tateishiet al., TGFBR1 and TGFBR2 were expressed in 32%and 18% of GC tissues, respectively, and TGFBR2expression was correlated with invasion and poorprognosis (Tateishi et al., 2000). Several studies onhuman GC cell lines have revealed that loss of TGFBR1expression and decreased expression of TGFBR2, orexpression of a truncated form of TGFBR2, may play animportant role in the inhibition of apoptosis (Yang et al.,1999; Zhuang et al., 1999; Park et al., 2001) and thathomozygous mutation of TGFBR2 is observed in two ofnine GC cell lines (unpublished data), although down-regulation of TGFBR1 and TGFBR2 in GC patients hasnot been investigated in detail. A homozygous mutationof TGFBR2, targeting a polyadenine tract in its codingsequence, could represent a replication-error (RER)-positive carcinoma, contributing to 10% of GC (Guo etal., 1998). Moreover, a paradoxical increase in TGFBsignaling has been shown in individuals with variouscongenital anomalies by heterozygous mutation inTGFBR1 or TGFBR2, and transgenic mice expressing adominant negative (kinase-domain deleted) TGFBR2(Denton et al., 2003; Loeys et al., 2005). Recently, aTGFBR1 inhibitor has been reported to prevent theinvasion and metastasis of SMAD4-deficient pancreaticcarcinoma cells (Subramanian et al., 2004), whichfurther suggests the significance of increased expressionof TGFBR1 and TGFBR2 in tumorigenesis.

Decreased expression of SMAD3 in 37.5% of GCsand restoration of TGFB responsiveness by introductionof SMAD3 into tumor cells has been reported by Han etal. (2004). These results are also consistent with the datapresented here (Table 5). Currently, we have confirmedthat SMAD4 down-regulation is critical for progressionof GC (Wang et al., 2007). Our data support thehypothesis that the alteration of TGFB signaling in GCis primarily generated not by TGFR2 inactivation, but bySMADs inactivation via various mechanisms (Wang etal., 2007) (Table 5).

The high level of CDKN1A expression in dysplasiaand the subsequent decrease in GC was reported by Xieet al. (Xie et al., 2004; Sun et al., 2005) (Table 5).Reduced expression of CDKN1B was reported in 76%of GCs without correlation with clinico-pathologicalparameters (Wiksten et al., 2002) (Table 5).Furthermore, Helicobacter pylori infection in CDKN1Bknockout mice was associated with significantly higherincidences of IM, dysplasia, and carcinoma in thestomach (Kuzushita et al., 2005). Kudo et al. reportedthat the expression of CDC25A was found in 38% ofgastric carcinomas without clinical significance (Kudo etal., 1997) (Table 5). The frequent expression of MYCand its relevance to advanced stages of carcinoma wasreported by Spandidos et al. (Spandidos et al., 1991)(Table 5).

Overexpression of TP53 in GC and its relevance to apoor prognosis has been reported in several papers(Okuyama et al., 2002; Xie et al., 2004; Al-Moundhri etal., 2005; Sun et al., 2005), the results of which areconsistent with our data here. The expression of RELAhas been reported in GC, although the clinicalsignificance of RELA in GC is controversial. Yamanakaet al. and Sasaki et al. reported that RELA expressionwas correlated with poor prognosis, whereas the resultsof Lee et al. indicate the exact opposite, i.e., RELAexpression was associated with better prognosis (Sasakiet al., 2001; Yamanaka et al., 2004; Lee et al., 2005).Our results are similar to those of Yamanaka et al. andSasaki et al. in that RELA expression was correlatedwith greater tumor size and was present at higher levelsin patient clusters with poorer prognoses.

The paradoxical function of TGFB signaling in GC

TGFB has the potential to function as a tumorsuppressor (via its effects on proliferation, replicationpotential, and apoptosis), or as a tumor promoter (via itseffects on migration, invasion, angiogenesis, and theimmune system). How can this dichotomy of function beresolved? Based on animal models and in vitro studies,Elliott and Blobe (2005) proposed a hypothesis thatduring early tumorigenesis, TGFB-mediated tumorsuppressor activity functions through a SMAD-dependent pathway, but that TGFB promoted tumorprogression later through a SMAD-independentpathway. However, no complete clinical study haspreviously supported this hypothesis. In the presentstudy, our data showing the up-regulation of TGFB,TGFBR1, TGFBR2, CDC25A, MYC, and the reducedexpression of SMAD1/2/3(N), SMAD2/3(N),SMAD4(C), SMAD4(N) in AGC (Fig. 6), is consistentwith Elliott’s hypothesis. Consequently, the prevalenceof SMAD-independent pathways due to the down-regulation of SMADs may be strongly pro-oncogenic(Fig. 5). TGFB-dependent down-regulation of MYC iscentral to its ability to inhibit proliferation in many cells(Chen et al., 2001). In addition, repression of MYC inTGFB signaling is known to result from the binding of aSMAD complex to the MYC promoter (Chen et al.,2001). Therefore, up-regulation of MYC in AGC may becentral to the functional disruption of the SMAD-dependent pathway (Fig. 5). At this point, it isnoteworthy that nuclear expression of SMAD1/2/3,SMAD2/3, and SMAD4 was reduced only in advancedstages (Fig. 6).

In conclusion, we have comprehensively andsystematically examined the expression of TGFBsignaling components, emphasizing the SMAD-dependent pathway in GC and its precursor lesions.There were extensive alterations in the expression ofTGFB signaling molecules and TP53 across the wholegastritis-dysplasia-carcinoma sequence, and our resultssuggest that the fundamental change in TGFB pathwaysignals from tumor suppressive to pro-oncogenic may be

1449


due to down-regulation of SMADs concomitant withtumor progression. These findings may be valuable inunderstanding the mechanisms of GC.

Acknowledgements. We are very grateful to Jung Sun Lee and MeeYoung Sim for their technical assistance, including sectioning andimmunohistochemistry, and for the financial support of the ResearchInstitute of Pharmaceutical Science. This study was partly supported bytwo grants of the National R&D Program for Cancer Control (A060353)and the Korea Health 21 R&D Project (A050260), Ministry of Health &Welfare, Korea.

References

Al-Moundhri M.S., Nirmala V., Al-Hadabi I., Al-Mawaly K., Burney I., Al-Nabhani M., Thomas V., Ganguly S.S. and Grant C. (2005). Theprognostic significance of p53, p27 kip1, p21 waf1, HER-2/neu, andKi67 proteins expression in gastric cancer: a clinicopathological andimmunohistochemical study of 121 Arab patients. J. Surg. Oncol. 91,243-252.

Chen C.R., Kang Y. and Massague J. (2001). Defective repression of c-myc in breast cancer cells: A loss at the core of the transforminggrowth factor beta growth arrest program. Proc. Natl. Acad. Sci.USA 98, 992-999.

Choi Y., Kim J.Y., Kwon M.J., Choi J.S., Kim T.J., Bae D.S., Koh S.S.,In Y.H., Park Y.W., kim S.H., Ahn G.H. and Shin Y.K. (2007).Expression profile of tight junction protein claudin 3 and claudin 4 inovarian serous adenomacarcinoma with prognostic correlation.Histol. Histopathol. 22, 1185-1195

Cordenonsi M., Dupont S., Maretto S., Insinga A., Imbriano C. andPiccolo S. (2003). Links between tumor suppressors: p53 is requiredfor TGF-beta gene responses by cooperating with Smads. Cell 113,301-314.

Cordenonsi M., Montagner M., Adorno M., Zacchigna L., Martello G.,Mamidi A., Soligo S., Dupont S. and Piccolo S. (2007). Integration ofTGF-{beta} and Ras/MAPK Signaling Through p53 Phosphorylation.Science 315, 840-843.

de Guise C., Lacerte A., Rafiei S., Reynaud R., Roy M., Brue T. andLebrun J.J. (2006). Activin inhibits the human Pit-1 gene promoterthrough the p38 kinase pathway in a Smad-independent manner.Endocrinology 147, 4351-4362.

Denton C.P., Zheng B., Evans L.A., Shi-wen X., Ong V.H., Fisher I.,Lazaridis K., Abraham D.J., Black C.M. and de Crombrugghe B.(2003). Fibroblast-specific expression of a kinase-deficient type IItransforming growth factor beta (TGFbeta) receptor leads toparadoxical activation of TGFbeta signaling pathways with fibrosis intransgenic mice. J. Biol. Chem. 278, 25109-25119.

Derynck R. and Zhang Y. (1996). Intracellular signalling: the mad way todo it. Curr. Biol. 6, 1226-1229.

Ebert M.P., Yu J., Miehlke S., Fei G., Lendeckel U., Ridwelski K., StolteM., Bayerdorffer E. and Malfertheiner P. (2000). Expression oftransforming growth factor beta-1 in gastric cancer and in the gastricmucosa of first-degree relatives of patients with gastric cancer. Br. J.Cancer 82, 1795-1800.

Elliott R.L. and Blobe G.C. (2005). Role of transforming growth factorbeta in human cancer. J. Clin. Oncol. 23, 2078-2093.

Fenoglio-Preiser C.M., Wang J., Stemmermann G.N. and Noffsinger A.(2003). TP53 and gastric carcinoma: a review. Hum. Mutat. 21, 258-

270.Genta R.M. and Rugge M. (2001). Review article: pre-neoplastic states

of the gastric mucosa--a practical approach for the perplexedclinician. Aliment. Pharmacol. Ther. 15 Suppl 1, 43-50.

Gold L. (1999). The role for transforming growth factor ß (TGF-ß) inhuman gastric cancer. Crit. Rev. Oncog. 10, 303-360.

Guo R.J., Wang Y., Kaneko E., Wang D.Y., Arai H., Hanai H.,Takenoshita S., Hagiwara K., Harris C.C. and Sugimura H. (1998).Analyses of mutation and loss of heterozygosity of codingsequences of the entire transforming growth factor beta type IIreceptor gene in sporadic human gastric cancer. Carcinogenesis 19,1539-1544.

Han S., Kim H.Y., Park K., Cho H.J., Lee M.S., Kim H.J. and Kim Y.D.(1999). c-Myc expression is related with cell proliferation andassociated with poor clinical outcome in human gastric cancer. J.Korean Med. Sci. 14, 526-530.

Han S.U., Kim H.T., Seong D.H., Kim Y.S., Park Y.S., Bang Y.J., YangH.K. and Kim S.J. (2004). Loss of the Smad3 expression increasessusceptibility to tumorigenicity in human gastric cancer. Oncogene23, 1333-1341.

Hara T., Ooi A., Kobayashi M., Mai M., Yanagihara K. and Nakanishi I.(1998). Amplification of c-myc, K-sam, and c-met in gastric cancers:detection by fluorescence in situ hybridization. Lab. Invest. 78, 1143-1153.

Ho S.B. (1996). Premalignant lesions of the stomach. Semin.Gastrointest Dis. 7, 61-73.

Ito M., Yasui W., Nakayama H., Yokozaki H., Ito H. and Tahara E.(1992). Reduced levels of transforming growth factor-beta type Ireceptor in human gastric carcinomas. Jpn. J. Cancer Res. 83, 86-92.

Jang S.J., Ahn M.J., Paik S.S., Kong G., Keum J.S., Park Y.W. and LeeJ.D. (1998). Expression of cyclin dependent kinase inhibitorp21WAF1 alone and in combination with p27KIP1 shows prognosticvalue in gastric carcinoma. J. Korean Med. Sci. 13, 369-376.

Kai T., Taketazu F., Kawakami M., Shimanuki K., Yamada S., MiyazonoK., Kato M. and Miyata M. (1996). Distribution of transforminggrowth factor-beta and its receptors in gastric carcinoma tissue. Jpn.J. Cancer Res. 87, 296-304.

Kang S.H., Bang Y.J., Im Y.H., Yang H.K., Lee D.A., Lee H.Y., LeeH.S., Kim N.K. and Kim S.J. (1999). Transcriptional repression of thetransforming growth factor-beta type I receptor gene by DNAmethylation results in the development of TGF-beta resistance inhuman gastric cancer. Oncogene 18, 7280-7286.

Kim S.J., Im Y.H., Markowitz S.D. and Bang Y.J. (2000). Molecularmechanisms of inactivation of TGF-beta receptors duringcarcinogenesis. Cytokine Growth Factor Rev. 11, 159-168.

Kim Y.H., Lee H.S., Lee H.J., Hur K., Kim W.H., Bang Y.J., Kim S.J.,Lee K.U., Choe K.J. and Yang H.K. (2004). Prognostic significanceof the expression of Smad4 and Smad7 in human gastriccarcinomas. Ann. Oncol. 15, 574-580.

Kozma L., Kiss I., Hajdu J., Szentkereszty Z., Szakall S. and Ember I.(2001). C-myc amplification and cluster analysis in human gastriccarcinoma. Anticancer Res. 21, 707-710.

Kudo Y., Yasui W., Ue T., Yamamoto S., Yokozaki H., Nikai H. andTahara E. (1997). Overexpression of cyclin-dependent kinase-activating CDC25B phosphatase in human gastric carcinomas. Jpn.J. Cancer Res. 88, 947-952.

Kuzushita N., Rogers A.B., Monti N.A., Whary M.T., Park M.J., AswadB.I., Shirin H., Koff A., Eguchi H. and Moss S.F. (2005). p27kip1

1450


deficiency confers susceptibility to gastric carcinogenesis inHelicobacter pylori-infected mice. Gastroenterology 129, 1544-1556.

Lee B.L., Lee H.S., Jung J., Cho S.J., Chung H.Y., Kim W.H., Jin Y.W.,Kim C.S. and Nam S.Y. (2005). Nuclear factor-kappaB activationcorrelates with better prognosis and Akt activation in human gastriccancer. Clin. Cancer Res. 11, 2518-2525.

Li X., Zhang Y.Y., Wang Q. and Fu S.B. (2005). Association betweenendogenous gene expression and growth regulation induced byTGF-beta1 in human gastric cancer cells. World J. Gastroenterol.11, 61-68.

Loeys B.L., Chen J., Neptune E.R., Judge D.P., Podowski M., Holm T.,Meyers J., Leitch C.C., Katsanis N., Sharifi N., Xu F.L., Myers L.A.,Spevak P.J., Cameron D.E., De Backer J., Hellemans J., Chen Y.,Davis E.C., Webb C.L., Kress W., Coucke P., Rifkin D.B., De PaepeA.M. and Dietz H.C. (2005). A syndrome of altered cardiovascular,craniofacial, neurocognitive and skeletal development caused bymutations in TGFBR1 or TGFBR2. Nat Genet. 37, 275-281.

Maehara Y., Kakeji Y., Kabashima A., Emi Y., Watanabe A., AkazawaK., Baba H., Kohnoe S. and Sugimachi K. (1999). Role oftransforming growth factor-beta 1 in invasion and metastasis ingastric carcinoma. J. Clin. Oncol. 17, 607-614.

Massague J. (1996). TGFbeta signaling: receptors, transducers, andMad proteins. Cell. 85, 947-950.

Massague J. (1998). TGF-beta signal transduction. Annu. Rev.Biochem. 67, 753-791.

Naef M., Ishiwata T., Friess H., Buchler M.W., Gold L.I. and Korc M.(1997). Differential localization of transforming growth factor-betaisoforms in human gastric mucosa and overexpression in gastriccarcinoma. Int. J. Cancer 71, 131-137.

Okuyama T., Maehara Y., Kabashima A., Takahashi I., Kakeji Y. andSugimachi K. (2002). Combined evaluation of expressions of p53and p21 proteins as prognostic factors for patients with gastriccarcinoma. Oncology 63, 353-361.

Onoda N., Maeda K., Chung Y.S., Yano Y., Matsui-Yuasa I., Otani S.and Sowa M. (1996). Overexpression of c-myc messenger RNA inprimary and metastatic lesions of carcinoma of the stomach. J. Am.Coll. Surg. 182, 55-59.

Pardali K., Kurisaki A., Moren A., ten Dijke P., Kardassis D. andMoustakas A. (2000). Role of Smad proteins and transcription factorSp1 in p21(Waf1/Cip1) regulation by transforming growth factor-beta. J. Biol. Chem. 275, 29244-29256.

Park S.H., Kim Y.S., Park B.K., Hougaard S. and Kim S.J. (2001).Sequence-specific enhancer binding protein is responsible for thedifferential expression of ERT/ESX/ELF-3/ESE-1/jen gene in humangastric cancer cell lines: Implication for the loss of TGF-beta type IIreceptor expression. Oncogene 20, 1235-1245.

Park Y.E., Choi K.C. and Choi Y.H. (1998). p21 expression andmutation in gastric carcinoma: analysis by immunohistochemistryand PCR-SSCP. J. Korean Med. Sci. 13, 507-512.

Pinto M., Oliveira C., Cirnes L., Carlos Machado J., Ramires M.,Nogueira A., Carneiro F. and Seruca R. (2003). Promotermethylation of TGFbeta receptor I and mutation of TGFbeta receptorII are frequent events in MSI sporadic gastric carcinomas. J. Pathol.200, 32-38.

Saito H., Tsujitani S., Oka S., Kondo A., Ikeguchi M., Maeta M. andKaibara N. (1999). The expression of transforming growth factor-beta1 is significantly correlated with the expression of vascularendothelial growth factor and poor prognosis of patients withadvanced gastric carcinoma. Cancer 86, 1455-1462.

Sandhu C., Garbe J., Bhattacharya N., Daksis J., Pan C.H., Yaswen P.,Koh J., Slingerland J.M. and Stampfer M.R. (1997). Transforminggrowth factor beta stabil izes p15INK4B protein, increasesp15INK4B-cdk4 complexes, and inhibits cyclin D1-cdk4 associationin human mammary epithelial cells. Mol. Cell Biol. 17, 2458-2467.

Sasaki N., Morisaki T., Hashizume K., Yao T., Tsuneyoshi M., NoshiroH., Nakamura K., Yamanaka T., Uchiyama A., Tanaka M. andKatano M. (2001). Nuclear factor-kappaB p65 (RelA) transcriptionfactor is constitutively activated in human gastric carcinoma tissue.Clin. Cancer Res. 7, 4136-4142.

Shin J.Y., Kim H.S., Lee K.S., Kim J., Park J.B., Won M.H., Chae S.W.,Choi Y.H., Choi K.C., Park Y.E. and Lee J.Y. (2000). Mutation andexpression of the p27KIP1 and p57KIP2 genes in human gastriccancer. Exp. Mol. Med. 32, 79-83.

Shitara Y., Yokozaki H., Yasui W., Takenoshita S., Kuwano H.,Nagamachi Y. and Tahara E. (1999). No mutations of the Smad2gene in human sporadic gastric carcinomas. Jpn. J. Clin Oncol. 29,3-7.

Spandidos D.A., Karayiannis M., Yiagnisis M., Papadimitriou K. andField J.K. (1991). Immunohistochemical analysis of the expressionof the c-myc oncoprotein in human stomach cancers. Digestion 50,127-134.

Subramanian G., Schwarz R.E., Higgins L., McEnroe G., ChakravartyS., Dugar S. and Reiss M. (2004). Targeting endogenoustransforming growth factor beta receptor signaling in SMAD4-deficient human pancreatic carcinoma cells inhibits their invasivephenotype. Cancer Res. 64, 5200-5211.

Sun Y., Li J.Y., He J.S., Zhou L.X. and Chen K. (2005). Tissuemicroarray analysis of multiple gene expression in intestinalmetaplasia, dysplasia and carcinoma of the stomach.Histopathology 46, 505-514.

Takeno S., Wirtz H.C., Lickvers K., Noguchi T., Scheven M., Willers R.,Gabbert H.E. and Mueller W. (2002). Transforming growth factorbeta type II receptor expression in gastric cancer: evidence for twoindependent subgroups. Anticancer Res. 22, 2247-2252.

Tateishi M., Kusaba I., Masuda H., Tanaka T., Matsumata T. andSugimachi K. (2000). The progression of invasiveness regarding therole of transforming growth factor beta receptor type II in gastriccancer. Eur. J. Surg. Oncol. 26, 377-380.

Wakefield L.M. and Roberts A.B. (2002). TGF-beta signaling: positiveand negative effects on tumorigenesis. Curr. Opin. Genet. Dev. 12,22-29.

Wang L.H., Kim S.H., Lee J.H., Choi Y.L., Kim Y.C., Park T.S., HongY.C., Wu C.F. and Shin Y.K. (2007). Inactivation of SMAD4 tumorsuppressor gene during gastric carcinoma progression. Clin. CancerRes. 13, 102-110.

Wiksten J.P., Lundin J., Nordling S., Kokkola A., von Boguslawski K.and Haglund C. (2002). The prognostic value of p27 in gastriccancer. Oncology 63, 180-184.

Xiangming C., Natsugoe S., Takao S., Hokita S., Ishigami S., TanabeG., Baba M., Kuroshima K. and Aikou T. (2001). Preserved Smad4expression in the transforming growth factor beta signaling pathwayis a favorable prognostic factor in patients with advanced gastriccancer. Clin. Cancer Res. 7, 277-282.

Xie H.L., Su Q., He X.S., Liang X.Q., Zhou J.G., Song Y. and Li Y.Q.(2004). Expression of p21(WAF1) and p53 and polymorphism ofp21(WAF1) gene in gastric carcinoma. World J. Gastroenterol. 10,1125-1131.

Yamanaka N., Sasaki N., Tasaki A., Nakashima H., Kubo M., Morisaki

1451


T., Noshiro H., Yao T., Tsuneyoshi M., Tanaka M. and Katano M.(2004). Nuclear factor-kappaB p65 is a prognostic indicator ingastric carcinoma. Anticancer Res. 24, 1071-1075.

Yang H.K., Kang S.H., Kim Y.S., Won K., Bang Y.J. and Kim S.J.(1999). Truncation of the TGF-beta type II receptor gene results ininsensitivity to TGF-beta in human gastric cancer cells. Oncogene18, 2213-2219.

Zhuang Z., Chen Y.L., Wang C.D. and Chen Y.G. (1999). Theexpression of TGFß and TGFß receptor I in gastric carcinoma and

precancerous lesions. Zhongguo Shiyong Neike Zazhi. 21, 342-344.

Zolota V., Batistatou A., Tsamandas A.C., Iliopoulos G., Scopa C.D. andBonikos D.S. (2002). Immunohistochemical expression of TGF-beta1, p21WAF1, p53, Ki67, and angiogenesis in gastriccarcinomas: a clinicopathologic study. Int. J. Gastrointest. Cancer32, 83-89.

Accepted June 17, 2008

1452


Extensive alteration in the expression profiles of TGFB ... · carcinoma (GC) is chronic atrophic...

Documents

Transcript of Extensive alteration in the expression profiles of TGFB ... · carcinoma (GC) is chronic atrophic...